U-Mart Project: Learning Economic Principles from the Bottom by Both Human and Software Agents

Authors:
Hiroshi Sato;Hiroyuki Matsui;Isao Ono;Hajime Kita;Takao Terano;Hiroshi Deguchi;Yoshinori Shiozawa
Affiliations:
-;-;-;-;-;-;-
Venue:
Proceedings of the Joint JSAI 2001 Workshop on New Frontiers in Artificial Intelligence
Year:
2001

Citing 0
Cited 1

A Multi-objective Genetic Algorithm Approach to Construction of Trading Agents for Artificial Market Study

Proceedings of the Joint JSAI 2001 Workshop on New Frontiers in Artificial Intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

U-Mart is an interdisciplinary research program of agent-based artificial market. U-Mart proposes an open-type test bed to study trading strategies of agents, behavior of the market and their relationship. An experiment open to public (Pre U-Mart 2000) using the proposed system is held in August 2000. More than 40 software agents (computer programs for trading) from 11 teams participated in this experiment. This paper reports the outline of the experiment, the trading strategies of the participated agents and the results of the experiment. While Pre U-Mart 2000 treated only software agents, the U-Mart system is designed considering participation of the human players as well as the software agents. A gaming simulation by human using the U-Mart system held in Kyoto University is also introduced briefly.